Document Summarizer Pack

from document_summarizer_pack.tool import run

result = run(
    action="summarize_document",
    text="""Abstract: This study examines the impact of transformer-based architectures
on low-resource language translation. We evaluated mBERT, XLM-R, and a novel
fine-tuning approach on 12 African languages with fewer than 100,000 parallel
sentences each. Our method achieved BLEU score improvements of 4.2 to 8.7 points
over baseline models. The key innovation is a curriculum learning strategy that
gradually increases sentence complexity during fine-tuning. Results demonstrate
that pre-trained multilingual models can be effectively adapted for extremely
low-resource pairs when combined with strategic data ordering. We release our
fine-tuning toolkit and evaluation benchmarks as open source.""",
    format="structured",
    max_length=150
)

print(f"Title: {result['title']}")
print(f"Summary: {result['summary']}")
print(f"Key Points:")
for point in result["key_points"]:
    print(f"  - {point}")